Research articlesThe origin of modern patterns of continental diversity in Mauritiinae palms: the Neotropical museum and the Afrotropical graveyard
Abstract
While the latitudinal diversity gradient has received much attention, biodiversity and species richness also vary between continents across similar latitudes. Fossil information can be used to understand the evolutionary mechanisms that generated such variation between continents of similar latitudes. We integrated fossil data into a phylogenetic analysis of the Mauritiinae palms, whose extant diversity is restricted to the Neotropics, but extended across Africa and India during most of the Cenozoic. Mauritiinae diverged from its sister lineage Raphiinae ca 106 Ma. Using ancestral state estimation and a lineage through time analysis, we found that diversity arose globally during the late Cretaceous and Palaeocene across South America, Africa and India. The Palaeocene–Eocene transition (ca 56 Ma) marked the end of global Mauritiinae expansion, and the beginning of their decline in both Africa and India. Mauritiinae disappeared from the Indian subcontinent and Africa at the end of the Eocene and the Miocene, respectively. By contrast, Neotropical diversity steadily increased over the last 80 Myr. Taken together, our results suggest that the Neotropics functioned as a continental-scale refuge for Mauritiinae palms, where lineages survived and diversified while global climatic changes that drastically reduced rainforests led to their demise on other continents.
1. Introduction
The pattern of decreasing species richness from the equator to the poles has received a lot of attention (e.g. [1–3]). Evidence shows that differences across continents also exist, suggesting different processes are acting across similar latitudes. The Neotropics and Afrotropics provide a particularly interesting comparison with marked differences in current species richness and community composition [4]. Raven et al. [5] showed that the South American continent harbours more than twice as many plant species as the Afrotropical region, a pattern that applies to other groups of organisms such as mammals [6], birds [7], amphibians [8,9] and butterflies [10]. The opening of the south Atlantic Ocean during the Early Cretaceous (145–100.5 Ma) initiated the separation of South America from Africa, which previously constituted most of the Gondwana supercontinent. Studying ancient lineages that diversified throughout the Cretaceous and the Cenozoic while these two continents drifted away from each other provides a unique opportunity to understand the emergence of variation in continental biodiversity.
The palm clade Mauritiinae found in modern Neotropical rainforests most likely originated during the early phase of the opening of the south Atlantic Ocean [11]. The fossil record indicates that Mauritiinae already thrived in Africa during the Late Cretaceous (100–66 Ma), and at the Cretaceous-Palaeogene (K-Pg) transition period (66 Ma), Mauritiinae spread across South America, Africa and the Indian subcontinent. At some point, however, the evolutionary history of Mauritiinae on the different continents diverged, as Mauritiinae are nowadays only found in the Neotropical region, where five extant species are recognized. Not only did the group survive in the Neotropics but one species, Mauritia flexuosa became one of the most dominant tree species in Amazonia, with ca 1.5 billion individuals [12,13].
Here, we use the rich fossil pollen record available for Mauritiinae [11] to: (i) estimate the timing of origin of the different lineages in a phylogenetic framework that integrates both extinct and extant species, (ii) track the different Mauritiinae lineages through time and space, and (iii) infer the timing(s) at which diversity arose and declined across different continents. Our comprehensive integration of fossil and molecular data sheds further light on the rise of differences in biodiversity across continents.
2. Material and methods
(a) Fossil record
All extant and extinct taxa belonging to the genera Mauritia, Echidiporites, Grimsdalea, Lepidocaryum, Mauritiella and Mauritiidites are collectively referred to as Mauritiinae. We extracted fossil information for Mauritiinae pollen taxa from Bogotá-Ángel et al. [11]. The dataset consisted of the most reliable first (FAD) and last (LAD) appearance datums in every country Mauritiinae pollen data has been found (table 1; electronic supplementary material, table S1).
| tip | date (Ma) | references |
|---|---|---|
| Echiporites barbeitoensis | 56.03 | Muller et al. [14] |
| Grimsdalea magnaclavata | 66.00 | Salard-Cheboldaeff [15] |
| Grismdalea polygonalis | 61.60 | Ikegwuonu et al. [16] |
| Mauritiidites crassibaculatus | 83.60 | Salami [17], Edet & Nyong [18] |
| Mauritiidites crassiexinus | 72.10 | Eisawi & Schrank [19] |
| Mauritiidites franciscoi franciscoi | 66.00 | van der Hammen & García de Mutis [20], Raymer [21] |
| Mauritiidites franciscoi minutus | 66.00 | van der Hammen & García de Mutis [20], Raymer [21] |
| Mauritiidites franciscoi pachyexinatus | 66.00 | van der Hammen & García de Mutis [20], Sarmiento [22], Colmenares & Teran [23] |
| Mauritiidites lehmanii | 83.60 | Salard-Cheboldaeff [24], Salard-Cheboldaeff [25]. Salard-Cheboldaeff [15], Eisawi & Schrank [19] |
(b) Molecular data
Molecular data were available for five taxa out of the nine extant Mauritiinae. The dataset consisted of up to three gene fragments (2565 base pairs, electronic supplementary material, file S1A-B). Three taxa were added based on morphological characters: Lepidocaryum tenue var gracile, Mauritiella pumila ‘small’ and Mauritiella pumila ‘large’. We included two outgroups—Raphia palmapinus and Oncocalamus mannii.
(c) Morphological data
We used data from Bogotá-Ángel et al. [11] to score four discrete morphological characters from pollen samples, for both fossil and extant taxa (including outgroups). Morphological features, like the number of apertures, aperture type and ornamentation, were based on light microscopy, while exine surface characters were determined with scanning and transmission electron microscopy (electronic supplementary material, file S2).
(d) Tip-dated phylogenetic tree
We combined molecular and morphological information to infer a phylogeny containing both extant and extinct taxa using a fossilized birth–death process ([26–28]; table 1). Substitution models for the three gene fragments were set as unlinked. We linked a lognormal relaxed molecular clock across the three gene fragments and used a strict-clock for the morphological character partition. We set an exponential prior for the root age with an offset of 86.3 Ma, corresponding to the oldest record of Mauritiinae pollen known according to Bogotá-Ángel et al. [11] and a mean of 30 Ma (97.5% quantile = 197 Ma). We performed two independent analyses using Beast 2.5.2 [29]. Each analysis ran for 150 million generations, sampling every 15 000 generations (electronic supplementary material, file S3). Adequate mixing and convergence of the Markov chain Monte Carlo analyses were verified in Tracer 1.6.0 [30]. For each run, we removed the first half of samples as burn-in and combined them using LogCombiner 2.5.2 ([29], electronic supplementary material, file S4). A maximum clade credibility tree was generated using TreeAnnotator 2.5.2 ([29], electronic supplementary material, file S5). Additional tests using a strict molecular clock showed no significant change (electronic supplementary material, file S6).
(e) Historical biogeography
We estimated past ancestral ranges with the Lagrange implementation of the dispersal-extinction-cladogenesis (DEC) model (version 20130526, [31]) and data from the fossil record. We designed a biogeographic model by assigning all taxa to one or more of three biogeographic regions, based on current and past distributions for extant and extinct taxa, respectively: South America, Africa (including the Middle East) and India (electronic supplementary material, table S2). Possible ranges included: South America + Africa, Africa + India and South America + Africa + India. We excluded the disjunct range South America + India, which was not found in the fossil record and unlikely. We set three time periods with different dispersal probabilities: 0–23.03, 23.03–66 and 66–140 Ma to account for increasing distance between continents (electronic supplementary material, table S5). We performed the ancestral state estimations on 50 trees randomly sampled from Beast posterior distribution, after removing outgroups.
Even though the fossil record indicates that some taxa disappeared from different regions at different periods of time, such range changes cannot be included in the standard Lagrange implementation of the DEC model. Thus, we assigned each tip to the maximum extent of its range. However, for each fossil terminal branch, we compared the range estimation from DEC with the fossil record and refined the dispersal and extirpation events along branches according to fossil information (electronic supplementary material, S7).
3. Results
We determined the divergence between Mauritiinae (extant + extinct) and Raphiinae to date to ca 93.6 Ma (95% credible interval (CI): 87.4–123.8 Ma). According to this estimation, the most recent common ancestor of extant Mauritiinae was estimated around 47.7 Ma (95% CI: 28.5–71.8 Ma; figure 1). The relationships between lineages, including extant taxa, were poorly supported. This lack of support is most likely owing to insufficient morphological and molecular information and recent phylogenomic trees at larger phylogenetic scale have obtained better resolution [32,33]. We performed all analyses on a posterior distribution of trees to account for phylogenetic uncertainty. Globally, our phylogeny indicates that Mauritiinae diversified rapidly during the Late Cretaceous (100.5–66.0 Ma), after which diversification drastically slowed down (figure 2). By the end of the Palaeocene (56 Ma), global diversity stopped increasing, stabilizing at around 10 lineages (figure 2).
Figure 1. Geographical range evolution is inferred based on the maximum clade credibility tree. Pie charts indicate the relative state probabilities of ranges. Coloured lines above the extinct terminal branches indicate the interval between each taxon's first and last appearance datum in each region. Numbers at nodes denote the median node ages (in Ma) from Beast analyses. Coloured boxes at the tips represent all regions where each taxon has been found at least once in the fossil record, the input information for ancestral state estimations. Figure 2. Global and continental number of lineages through time. (a) Grey lines are lineages through time plots drawn from the posterior distribution of Beast analyses (50 trees). Orange, purple and green curves are areas of tropical forest estimated by Kissling et al. [34] for different epochs for each region. (b) Number of lineages in South America, Africa and India as inferred from biogeographic ancestral state estimation on 50 trees from the Beast posterior distribution. Arrows indicate the time at which taxon richness peaked in each region. (c) Number of dispersal events between regions inferred from the ancestral biogeographic estimation. (d) Relative probabilities of ranges at the root were inferred from a biogeographic ancestral state estimation performed on 50 trees. For each range, the first and second boxplot correspond to the relative probability at the left and right branch from the root, respectively.
The Palaeotropics were inferred as the most probable range for both descendent lineages of the crown node of Mauritiinae (mean relative probability = 0.42 and 0.57, respectively; figure 2). Dispersal events towards the Neotropics and Indian subcontinent occurred relatively rapidly, and diversity in the three regions started to rise during the Late Cretaceous (100.5–66.0 Ma; figures 1 and 2). Towards the end of the Late Cretaceous and Palaeocene (66.0–56.0 Ma), diversity in both Africa and Asia reached their highest point, with about 10 and three lineages, respectively (figure 2). Slowly rising during the late Cretaceous and Palaeocene, Neotropical diversity at the beginning of the Eocene (56.0 Ma) was comparable to the African Mauritiinae.
The Palaeocene–Eocene boundary (56 Ma) marks the breaking point between diversification in South America and the other regions. Diversification in Africa and India came to a halt and declined (figure 2) and by the end of the Eocene (33.9 Ma), Mauritiinae had already gone extinct in India. African diversity started declining after the Eocene–Oligocene boundary (33.9 Ma) and the last Mauritiinae lineages went extinct by the end of the Miocene (5.33 Ma).
The Cenozoic decline in diversity inferred for both Africa and India did not occur in South America. Diversity in South America increased during the Cenozoic to reach about 15 taxa during the Pliocene (5.3–2.6 Ma; figure 2). Four taxa went extinct during that period (including the three morphologically distinct Mauritiidites franciscoi fossil lineages). The linear increase of lineages through time in South America, however, indicates a slowing rate of diversification over time (figure 2).
4. Discussion
(a) Late Cretaceous–Palaeocene (100.5–56 Ma): Mauritiinae evolve and thrive under warm and wet climate
Results from our fossilized birth–death analysis indicate that Mauritiinae diverged from its sister group during the middle of the Cretaceous, much earlier than previously understood. Couvreur et al. [35] in their phylogeny of the palm family used a constraint of 65 Ma as a minimum age for the stem of Mauritiinae based on Mauritiidites fossils from the Maastrichtian (72.1–66.0 Ma, [36]). Bogotá-Ángel et al. [11] identified the appearance of fossils attributed to Mauritiinae between 94 and 83 Ma, which we used here to calibrate the stem of Mauritiinae. Our estimate for the divergence between Mauritiinae and Raphiinae has impacts at deeper time scales, since Bogotá-Ángel et al.'s [11] revision alone pushes the emergence of first Mauritiinae fossils almost to the age that Couvreur et al. [35] estimated for the crown of all palms (100 Ma, 95% highest posterior density 108–92), suggesting that the crown age of palms has been greatly underestimated.
Diversification of the Mauritiinae was initiated when the African and South American continents started to separate from each other during the Cretaceous. We inferred a most likely African origin, and dispersal towards other continents occurred not long after with the accumulation of lineages towards the end of the Cretaceous. At the K-Pg transition, Pan et al. [37] reported the disappearance of 16 (out of 34) palm or palm-like pollen species from the fossil record of west Africa, but Mauritiidites and Grimsdalea were two of the six genera that did not decline. We also find no signal of global extinction associated with the K-Pg mass extinction (66.0 Ma), which instead marks the peak of Mauritiinae diversification with lineages spreading across three continents and the highest number of taxa. The Late Cretaceous and Palaeogene were characterized by warm temperatures and ever-wet tropical biomes [37]. During the Palaeocene, palms became abundant in these environments, and Mauritiinae lineages such as Mauritiidites franciscoi typified tropical communities during that period [14,38,39]. This widespread wet tropical biome, also known as the Palmae Province [40,41], extended across much of South America, Africa and the Indian subcontinent (still isolated from Eurasia).
(b) Eocene–Miocene (56–5.3 Ma): Mauritiinae decline in Africa and India as climate cools
At the Palaeocene–Eocene thermal maximum (PETM, 55.5 Ma), tropical and subtropical-like forests covered most of South America [42] and their presence, even in Antarctica, at that time is widely evidenced by the fossil record [43]. After the PETM, and particularly at the Eocene–Oligocene Transition (EOT), global temperatures rapidly decreased [44]. This climate change is associated with the opening of the Southern Ocean gateways around Antarctica, which modified ocean circulation and prompted Antarctic glaciation [44]. Tropical and subtropical forests retreated towards lower latitudes [45], causing distinct trajectories across continents. African Mauritiinae started declining during the Eocene (56–33.9 Ma). While mangrove vegetation thrived during the late Palaeocene and early Eocene, with warm climate and the presence of the Sahara Sea in northern Africa [46], the persistence of a ‘pan-African rainforest’ spreading from east Africa to west Africa during the Eocene remains controversial (e.g. [47]). However, an important turnover event in vegetation has been documented at the EOT in Africa [24,46,48], with widespread fragmentation of rainforests and the spread of savannah and woodland biomes [47,49,50] as well as extinctions of palms [37].
We found that African Mauritiinae diversity was halved during the Oligocene (33.9–23.0 Ma) and despite the reappearance of warm and wet conditions during the first half of the Miocene, the decline continued. After the middle Miocene, climate cooling had an additional negative effect on palms and rainforests in general [51], with a slow replacement by a more arid climate, the expansion of savannahs, the appearance of the Sahara Desert (7 Ma, [52]) and the spread of C4 grass-dominated biomes [53,54]. At the end of the Miocene, the last Mauritiinae taxon (Grimsdalea polygonalis) disappeared from the African fossil record.
On the Indian subcontinent a decline of Mauritiinae also occurred at the EOT. After reaching its peak at the Cretaceous–Palaeogene transition period, Indian Mauritiinae went rapidly extinct from that region during the Eocene [11].
(c) Palaeocene—present (66–0 Ma): Mauritiinae find refuge in South America
In contrast with Africa and India, the number of Mauritiinae lineages in South America appears to have linearly increased throughout the Cenozoic. Palms are widespread nowadays and locally abundant in South American forests; a major difference compared to other tropical regions where palms are rarer [55]. Mauritiinae are abundant in South America, mostly in the Amazon drainage basin. Mauritia, Mauritiella and Lepidocaryum occur in various environments and are often associated with humid conditions such as swamps, mangroves, floodplains and river margins. South American rainforests suffered the least from the post-PETM global cooling (figure 2). Africa instead saw the greatest rainforest loss [34,43] and accompanying continental aridification. High extinction in Africa has long been proposed as a mechanism explaining the diversity- and endemism-poor rainforests (e.g. [46,56]) and the extant phylogenetic structure of African palm communities [34]. Baker & Couvreur [57], however, suggested that the high species diversity of Neotropical palms results from climatic and geological conditions promoting higher speciation rates than in Africa. Our results clearly emphasize the role of extinction rather than speciation in explaining the modern distribution of Mauritiinae. Climate change associated with African aridification and forest fragmentation probably led to the demise of African Mauritiinae. Instead, by maintaining suitable conditions for Mauritiinae, South America appears to have acted as a refuge for these ancient palms. Carvalho et al. [58] recently found that the end-Cretaceous events were pivotal in shaping South American rainforests. Key extinctions redefined the forest structure and diversity and may have enabled the diversification of new taxa, typically Mauritiinae, which are nowadays abundant in South America.
Data accessibility
Sequence information used for phylogenetic reconstruction is available on Genbank. Accession codes are provided in the electronic supplementary material [59]. Morphological characters used for phylogenetic reconstruction are available in the electronic supplementary material, as well as the trees generated for the analyses.
Authors' contributions
C.D.B.: conceptualization, data curation, formal analysis, funding acquisition, methodology, writing—original draft, writing—review and editing; D.S.: conceptualization, data curation, methodology, writing—review and editing; C.H.: conceptualization, data curation, writing—review and editing; G.B.-A.: data curation, writing—review and editing; A.A.: conceptualization, funding acquisition, writing—review and editing; N.C.: conceptualization, data curation, formal analysis, methodology, writing—original draft, writing—review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration
We declare we have no competing interests.
Funding
The research presented in this paper is a contribution to the strategic research area Biodiversity and Ecosystems in a Changing Climate, BECC, which supported N.C.'s research work. In addition, C.D.B., D.S. and A.A. are supported by the Swedish Research Council (2017-04980, 2019-04739 and 2019-05191 respectively); D.S. and A.A. are supported by MISTRA (BioPath); A.A. is also supported by a grant from the Kew Foundation at the Royal Botanic Gardens, Kew; and D.S. also received funding from the Swiss National Science Foundation (PCEFP3_187012).
Acknowledgements
We would like to thank Luis Palazzesi, the Editor and Associate Editor and two anonymous reviewers for their contribution to improving our manuscript. We would also like to thank Francine Almeida for submitting sequences to GenBank.
